Smerity

18 ms·

91.
by Smerity 10y ago
When you say "exactly these sorts of dynamic graphs", what do you mean? TensorFlow has support for dynamic length RNN unrolling but that really doesn't extend well to any dynamic graph structure such as recursive tree structu
92.
by Smerity 10y ago
Only a few months ago people saying that the deep learning library ecosystem was starting to stabilize. I never saw that as the case. The latest frontier for deep learning libraries is ensuring efficient support for dynamic computation grap
93.
by Smerity 10y ago
I'm glad this is happening and honestly surprised this didn't happen earlier. fchollet, the Keras main author, works at Google and had integrated TensorFlow into Keras very early on (given he'd known about the existence long
94.
by Smerity 10y ago
First, nawwwww, thanks :) Second though, I've sadly stopped responding (or even reading) many of the machine learning and especially broader artificial intelligence posts on Hacker News. The discussions are rarely discussions that one
95.
by Smerity 10y ago
To the Hacker News community, as I'm genuinely curious, does this link contribute anything to you or your understanding of deep learning? 29 people voted for it, zero comments, and it's still on the front page after five hours. It
96.
by Smerity 10y ago
See my reply to technics256 and feel free to ask other questions :)
97.
by Smerity 10y ago
For general introductory material in this style from Stanford, CS231n (fairly general but specialization in vision) and CS224d (specialization in DL for NLP) are great. The material for both of these are online for free and the video lectur
98.
by Smerity 10y ago
I work in the field and I'm with aaronjg - this is ancient in the scheme of deep learning and very far from modern best practices. I honestly find it confusing when I see links like this hit the top of the page with such a strong numbe
99.
by Smerity 10y ago
I'll note that my MS was hugely useful and didn't result in publications directly. The mileage of your MS or PhD is dependent on many factors. OpenAI and Google Brain, like most other more research driven deep learning institution
100.
by Smerity 10y ago
Key point given the current climate, if the Trump presidency adds a restrictive robots.txt to all .gov domains they will prevent the Internet Archive Wayback Machine from showing history on any of those domains. Not only would this oblitera
101.
by Smerity 10y ago
If people are interested in the underlying architecture of Google's Neural Machine Translation (GNMT) system, I wrote an article that builds it up piece by piece. While it's intended for people who are likely to implement GNMT or
102.
by Smerity 10y ago
Hey HN, I'm author of the article. If there's anything confusing, I'd be happy to help :) The GNMT architecture is likely to keep popping up in Google's papers due to the amount of engineering that's been dedicated
103.
by Smerity 10y ago
I provided a write-up of interesting ICLR submissions, primarily for NLP, but have also provided a small snippet explaining why I thought the paper was important or how it fit with other recent relevant papers. http://smerity.com
104.
by Smerity 10y ago
I'm not the author but can speak to some of this. Some amount of this was answered by the author himself in his blog post "Dead Code Should Be Buried – Why I Didn't Contribute to NLTK"[1]. While there are portions on NLT
105.
by Smerity 10y ago
The entire situation is downright despicable. If you listen to the podcast, they made employees stay in an active crime scene (bank robbery) where the thief defecated all over the floor to make phone calls to try to open new accounts. Many
106.
by Smerity 10y ago
I highly recommend listening to a recent episode[1] of Planet Money. This was by no means an accident. As a short summary of one employee's plight: Ashley, working for them and making only $35k per year in San Francisco, was continuall
107.
Introducing Explosion AI – Custom NLP Models by the Makers of SpaCy(explosion.ai)
3 points by Smerity 10y ago|0 comments
108.
by Smerity 10y ago
Have you seen the LibriSpeech ASR corpus[1]? It's large scale (1k hours) Creative Commons licensed English speech with the original book text. The data is derived from read audiobooks from the LibriVox project and has been carefully se
109.
by Smerity 10y ago
Not necessarily - they're listed as CC-BY on Flickr but that doesn't necessarily make it true. Specifically the user may not have authority to give out such copyright (as I assume for any case where the user "accidentally&quo
110.
by Smerity 10y ago
This is the tactic used by the CNN / Dailymail dataset[1] released by Google Deepmind. In that situation you want everyone to have the same and original file. The contents of the URL may be updated and/or disappear. The original U
111.
by Smerity 10y ago
$0.9 per K80 GPU per hour, while expensive, opens up so many opportunities - especially when you can get a properly connected machine. Just as an example of the change this entails for deep learning, the recent "Exploring the Limits of
112.
by Smerity 10y ago
If anyone is interested in exploring this, I highly recommend skipping to the blackjack exercise. This isn't to say you should read the book fully but I feel blackjack is simple and enough to get you addicted. Starting with Monte Carlo
113.
by Smerity 10y ago
Microsoft Azure's K80 based N-series are the best that I've seen so far for by the hour instances. They're currently in preview availability[1]. Compared to the best AWS GPU, the Azure K80s are (a) cheaper and (b) many times
114.
by Smerity 10y ago
This article lacks any real evidence they're ahead though, which is essentially equivalent to "behind" if (a) the majority of the advances are democratized by your competitors and (b) you lack the ability to attract top tier
115.
by Smerity 10y ago
None of this really answers the overlying question that Jerry Kaplan and Oren Etzioni raised. The question raised by most in the field isn't whether Apple use AI/ML internally, the real question is why they avoid the research comm
116.
A Glimpse at the Internet Ecosystem Through the Lens of the Netflix CDN(arxiv.org)
1 points by Smerity 10y ago|0 comments
117.
Bandwidth Costs Around the World(blog.cloudflare.com)
6 points by Smerity 10y ago|1 comments
118.
Correcting Intel's Deep Learning Benchmark Mistakes(blogs.nvidia.com)
141 points by Smerity 10y ago|37 comments
119.
by Smerity 10y ago
I may be wrong but from what I read the solution for Healthcare.gov didn't involve a primary contractor, it involved changing policies to reflect start-ups[1]. The Healthcare.gov debacle and resulting "start-up culture" resul
120.
by Smerity 10y ago
An interesting thing to note that is that Australia selected IBM to run the digital Census (which, surprise, failed horribly) whilst one of their states, Queensland, has an explicit IBM ban due to an ongoing $1.2 billion dollar contract deb

More ›